1:1 mentoring with Big Tech AI engineers
LLM & Agentic

Other Agent Frameworks

Translate what you know into the OpenAI Agents SDK, Google ADK and CrewAI: how each stops its loop, hands off between agents, pauses for approval, survives a crash and reaches Claude, with the defaults that bite.

Last updated

Production11 min readFirst readYour First Agent SDK AgentProductionising LangGraph

After this section you can

  • Map an unfamiliar agent framework onto five questions: stop, hand off, approve, survive a crash, reach the model
  • Read and adapt agent code in the OpenAI Agents SDK, Google ADK and CrewAI
  • Name the defaults that bite in each: loop caps, tracing, model routing, replayed tools
44

Other Agent Frameworks: OpenAI Agents SDK, Google ADK, CrewAI

The customer already picked a framework, and it is not the one you know. Every agent framework runs the same loop; what changes is the vocabulary and the defaults. Learn to translate, and to find the defaults that bite.

Key idea

Under every framework is the loop from The Agent Loop: call the model, run the tools it asks for, repeat until it answers. To learn a new framework, answer five questions: how the loop stops, how one agent hands work to another, where a person approves, what survives a crash, and how it reaches your model. The answers differ more than the names suggest.

One loop, five default caps
THE SAME LOOP EVERYWHERE model call wants tools, or answers tool calls your code runs them asks for tools results go back in final answer the loop stops or the cap is hit error, or a forced answer THE DEFAULT CAP, BY FRAMEWORK OpenAI Agents SDK 10 turns max_turns then MaxTurnsExceeded Google ADK 500 LLM calls per run max_llm_calls set in RunConfig CrewAI 25 iterations max_iter the docs say 20 LangGraph 10,007 super-steps GraphRecursionError raised at the cap Claude Agent SDK none no turn cap max_turns unset cap turns or dollars The defaults differ a thousandfold and count different things. Set the cap yourself, in the unit you can afford.

Related

More in LLM & Agentic

Tokens, Context Windows & Sampling

Preview

What a language model actually does on every call: split text into tokens, predict one next token at a time, fit everything into a fixed context window, and pick each token by sampling — with interactive temperature and context-budget explorers.

How LLMs Are Built

Preview

Complete lifecycle of large language models from pre-training through fine-tuning, RLHF, and deployment — with architecture diagrams and production considerations.

How LLMs Call Tools

Preview

How LLMs use function calling and tool use — the mechanics behind tool-calling agents, from prompt engineering to structured output.

Prompt Engineering Foundations

Preview

Where an instruction belongs — tool description, system prompt, or user turn — why emphasis that worked on older models now causes over-triggering, and the prompt patterns that have expired.

Get full access to all 149 sections with code examples, diagrams, and interactive animations.

Unlock Premium